Visual Statistics Models Instantly in Memory

Before in-memory analytics, the process of building models, testing, revising, and finally deploying them could take weeks or more per model.

With today's tools, however, analysts can test theories, tweak models, and generate visualizations against large datasets in minutes. It's hard to grasp the scale of the computational gymnastics going on behind the scenes, because the whole process looks easy.

In the video below, SAS data scientist Wayne Thompson demonstrates SAS Visual Statistics, a suite that makes advanced analytics operations as effortless as pointing and clicking. We got a sneak peek back in January; the software was released to the general public today.

According to a press release, Visual Statistics is designed to work with Hadoop and other databases, running on specialized appliances or commodity hardware. The improvements in productivity, accuracy, and speed have proven game-changing for early adopters. From the release:

With SAS Visual Statistics, multiple users can build and modify predictive models on large volumes of diverse data, using analytical methods that include regression and estimation, classification, and clustering. They can then present the analytical results visually, which speeds the ability to discover relationships between variables and determine which will positively affect outcomes.

In-memory processing keeps the entire dataset in system memory instead of writing anything to disk, which, as you can see in the video, enables analysts to execute tasks with unprecedented speed. The drag-and-drop interface is already familiar to professionals who use SAS Visual Analytics, which the company says is already licensed at more than1,500 customer sites.

SAS Visual Statistics enables analysts to compare multiple linear regression models and improve accuracy.
SAS Visual Statistics enables analysts to compare multiple linear regression models and improve accuracy.

One early adopter, Dutch debt management firm DirectPay, has been using Visual Statistics in conjunction with Visual Analytics to gain deeper insight and communicate it more efficiently across the organization.

As explained in a case study last year, DirectPay works in the delicate field of debt collection, and accurate modeling of customer data can mean the difference between successful operations and angry, delinquent debtors. SAS tools enable fast and accurate analytics plus easy sharing via tablets and other mobile devices.

Colin Nugteren, operations manager at DirectPay, seems enthused about Visual Statistics: "Its speed and usability make it a great tool. We are certain that this new software will improve both model quality and model deployment time, since more variables can be tested in different combinations in a short amount of time."

SAS is quick to point out that Visual Statistics isn't just for large enterprises and large datasets. "It is equally valuable for departmental deployments or midsize businesses, operating on several platforms, including database appliances from Pivotal and Teradata, Hadoop distributions from Cloudera and Hortonworks, or departmental servers," the release states.

Members, do you use Visual Analytics currently, and will you try Visual Statistics? What do you find appealing about these new tools? Does the feature set make it worthwhile to upgrade? Share your impressions below.

— Michael Steinhart, Circle me on Google+ Follow me on TwitterVisit my LinkedIn pageFriend me on Facebook, Executive Editor,

Related posts:

Michael Steinhart, Contributing Editor

Michael Steinhart has been covering IT and business computing for 15 years, tracking the rising popularity of virtualization, unified fabric, high-performance computing, and cloud infrastructures. He is editor of The Enterprise Cloud Site, which won the Least Imaginative Site Name award in 2012, and he managed, a community of IT professionals taking their first steps into cloud computing. From 2006 to 2012, Steinhart worked as an executive editor at Ziff Davis Enterprise, writing and managing research reports, whitepapers, case studies, magazine features, e-newsletters, blog posts, online videos, and podcasts. He also moderated and presented in dozens of webinars and virtual tradeshows. He got his start in IT journalism at CMP Media back in 1998, then moved to PC Magazine, managing the popular Solutions section and then covering business technology and consumer software. He holds a Bachelor of Arts degree in communications/journalism from Ramapo College of New Jersey.

Biosensors, Robots Replacing Human Guards

In prisons and other security settings, electronic devices are predicting and detecting anomalous behavior with high accuracy and reducing the need for human personnel.

Questioning the Ethics of Learning Analytics

Schools around the country are weighing the benefits and ethical questions of learning analytics.

Visual Statistics - Thanks from SAS
  • 5/16/2014 5:30:13 PM

Appreciate several of the nice comments.  We just completed interactive in-memory decision trees and are about to rap up k-means clustering for our release.  You can also create derived new columns for the leaf ids or the cluster ids and with the click of a mouse add these as group by variables in another model; stratifed modeling along the model factory theme.  Price is based on the size of the server and we are bullish on Hadoop.  Michael happy to do an updated demo anytime. Just give the SAS folks a shout.   Thanks again all.  Wayne Thompson

Re: Awesome
  • 5/15/2014 10:21:31 AM

So maybe "business necessity" wasn't the best word choice as, you're right, certainly many analytics shops would find the capabilities fitting of their needs. Really what I was angling at was the cost -- does that gentleman's desire for a model factory and automated diagnostics for models translate into his ability to justify the software spend? I'm not saying this to suggest that Visual Statistics is pricey, or too pricey, but to point out that cost is a factor. How much, by the way, does Visual Statistics cost? Any info to share on price? 

Re: Awesome
  • 5/15/2014 7:51:29 AM

You've done a great job with the demo Micheal and I would really like to get my hands on the software. This looks so easy. Imagine being able to do all that within seconds. Really awesome.

Re: Awesome
  • 5/14/2014 10:22:42 PM

When you talk about business necessity, though, it seems as though the features in VS really are meeting genuine enterprise needs. One of the speakers at SAS Health Analytics Executive Forum today specifically talked about his wish for a model factory and automated diagnostics for models. Someone in the audience pointed out that VS delivers these things, and he was surprised to hear it. 

Re: Awesome
  • 5/14/2014 1:45:20 PM

I'm with you there, Jeff. I love when a little personality is sprinkled into the mix, especially when the product is still high quality.

Re: Awesome
  • 5/14/2014 12:52:04 PM

I attended SAS Global Forum and Executive Conference a couple months back, and one of the "opening acts" at the kickoff keynote gala was a demo of Visual Statistics. It really does make you sit up, take notice, and say, "Wow!" Of course, seeing something really cool doesn't always translate into a business necessity and tool purchase, but I would expect that many statisticians/analytics professionals who see this in action will want to play around with it some.

Re: Awesome
  • 5/13/2014 10:07:32 PM

I think his enthusiasm is infectious, Jeff. He's clearly excited by the capabilities of the software and by the ways it supports traditional and next-generation workstyles.

When we shot the demo, Wayne ran through the whole thing twice -- once in the medium shot and once so we could have close-up views of the screens.  I took notes frantically so we could match the footage correctly. 

  • 5/13/2014 9:38:22 PM

Very awesome.  Wayne says he's an 'old school data scientist' and 'that's how I like to roll' and 'automagically'  too funny, I was cracking up.  But all that aside the tools look impressive.


Re: Impressive model production
  • 5/13/2014 6:01:46 PM

As a lay person, I find it interesting -- and impressive -- but I don't necessarily appreciate it at the same level as someone who actually knows what's happening behind the scenes. Charitable donations are the go-to demo scenario for SAS's analytics products, but you're right that the tools would be very useful for projects that model human activity on a large scale.

Impressive model production
  • 5/13/2014 5:56:31 PM


Very impressive demo, especially the drag-and-drop capability to produce predictive models in what looks like seconds.

In addition to the traditional private sector, I could see possible applications in public-sector services (like public transportation) and political campaign operations.